Strategies for robust, accurate, and generalizable benchmarking of drug discovery platforms
2024

Improving Drug Discovery Benchmarking

publication Evidence: moderate

Author Information

Author(s): Van Norden Melissa, Mangione William, Falls Zackary, Samudrala Ram

Primary Institution: Cold Spring Harbor Laboratory

Hypothesis

Can optimizing benchmarking protocols improve the performance of drug discovery platforms?

Conclusion

The study found that updated benchmarking protocols significantly improved the accuracy of drug candidate predictions.

Supporting Evidence

  • CANDO ranked 7.4% of known drugs in the top 10 compounds for their respective diseases based on CTD mappings.
  • This ranking increased to 12.1% with mappings from the Therapeutic Targets Database.
  • Performance was weakly correlated with indication size and moderately correlated with compound chemical similarity.
  • A new benchmarking protocol, compbench, was created for consistent assessment of drug discovery platforms.

Takeaway

This study shows how better testing methods can help find the best medicines for diseases.

Methodology

The study revised existing benchmarking protocols and optimized parameters in the CANDO platform to assess drug candidate predictions.

Digital Object Identifier (DOI)

10.1101/2024.12.10.627863

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication